Vote Operator for different attributes
I thought of using a remove attribute operator before each learner.
Will the resulting model (taken from the "mod" output of the vote operator) account for all the predictors?
What confuses me is the way the first learner is used. As I understand it, the predictions of all other learners are used as input to the first one. What should be used as the first learner to achieve normal voting (ie output the majority class in a classification problem)?
Also, since it is a simillar subject, how can I bias the model selection for each class
if model A predicts class 1 then output is class 1
else output is model B prediction